System reconfiguration to avoid deadlocks in automated manufacturing systems

Recent demands for small and medium-batch production runs have put additional constraints on automated manufacturing systems. Controlling the processes in such systems is difficult due to concurrent and asynchronous interactions that cause system deadlocks. Deadlocks occur in a system when a set of concurrent processes becomes interlocked in such a way that they cannot be completed. This paper reports on the control of deadlocks in a simple automated manufacturing system consisting of three workstations arranged around a transporter and process a single part. Deadlock detection and avoidance were performed off-line and the system was reconfigured by re-allocating the buffer capacity at the critical workstations. Petri Net formalism was used as the modeling tool for its ability to model concurrent and asynchronous activities, and its sound mathematical foundations for analysis. The model was numerically tested on the SUN Workstation under varying operating characteristics to study deadlock-free performance and compute effective buffer capacity at the critical workstations.